Given that the range of both hardware and software which could be incorporated into a marine fishery resources GIS is extremely large, it is important to mention how our discussion of this will necessarily be limited. Firstly, we shall only discuss hardware and software which is directly relevant in the sense that it would be both very useful to a marine fisheries GIS and that it could be considered as being within the working sphere of most GIS operators. Secondly, detail will not be given on how any piece of hardware actually works, i.e. our concern will be primarily with its usefulness, any limitations and its performance. Finally, it will not be possible to consider a whole range of different models of specific hardware items - we will restrict ourselves to “core items”. Readers who wish to obtain further information on hardware for GIS's could refer to Carstensen and Campbell (1991) or Faust et al (1991), plus specialised hardware and peripherals trade magazines. Further information on software can be obtained from Cambridge Market Intelligence (1993) and some of the other trade sources shown in Chapter 8.
It is pertinent to mention that the substance of most of this chapter is likely to date quickly. Technological advances in almost all areas of computing are such that new types of hardware, refined models of existing hardware and falling prices, are all occurring at a rate which could make much of this text obsolete in two to three years. As Frank et al (1991) have noted,
“There are no indications that the speed of this development will decrease in this decade. Even though there are ultimately physical laws that will limit speed and miniaturization, such as the speed of light and the need of at least one electron to store one bit of data, these limits are far beyond current technology.”
A similar revolution is also occurring in software development, with the major software houses constantly bringing out new versions of existing packages. To exemplify this rapid rate of GIS technological advancement, Table 5.1 shows that we are already in the so-called 5th generation of GIS's.
Throughout this section we will attempt to give indications of current purchase prices (in US$) of typical equipment or software described. Clearly these are only indicative and they will be subject to variations for a variety of reasons. Additionally, it will be clear that there will always be extra costs to consider such as peripherals, consumables, servicing, operator time, insurance, depreciation, etc. It is also important to remember that, for most hardware items, there are many manufacturers and that there are not yet any standards with regard to ensuring compatibility between hard and software. So, if buying a plotter for instance, where Hewlett-Packard (HP) are one of the leading makers, then make certain that your plotter includes drivers which emulate the HP. Although we can advise on plotter and printer prices, before purchasing a particular hardware item the buyer should also enquire about the output price per hard copy sheet.
Table5.1 Characteristics of the GIS Eras (as recognised by Antenucci, 1993)
|Circa||Document Platform||Generation||Processor Configuration||Database Configuration||Data Model Graphic||Data Model Attributes|
|1968–1975||Mainframe||1||Centralized||Centralized||Raster||Flat file/ hierarchical|
|1975–1982||Mini- computers||2||Centralized||Centralized||Raster||Hierarchical -relational|
|1982–1988||Workstations/ PCs||3||Partially||Centralized distributed||Vector||Relational|
|1988–1992||Workstations/ PCs||4||Distributed||Partially distributed||Vector||Relational|
|1993||Servers/ Workstations/ PCs||5||Distributed||Distributed||Vector/ Raster||Relational|
We should initially remind readers that in Chapter 2 many of the hardware items which are associated with data gathering have already been discussed, and in Chapter 4 we described the essential hardware which are used for GIS data capture and inputs, i.e. digitisers and scanners. This means that the hardware discussed here will be that associated with computer processing, data storage and the various means of obtaining GIS output. Some points to consider before purchasing any hardware are discussed in section 188.8.131.52.
Many readers will aware that the processing hardware (the computers themselves) have conventionally been classified under headings which broadly reflect their processing power. Thus processors range downwards from mainframes to minicomputers, to workstations, to micro or personal computers (PC's) and finally to lap-tops or notebooks (Table 5.2). We will not undertake a separate description of lap-tops or notebooks, since apart from their portability and mains free independence, their specifications and functionality are virtually the same as micro- computers. Originally GIS software was designed for use on the larger processors but recently there has been a marked movement towards supplying software which is specifically aimed at the smaller processors. Thus Maguire and Dangermond (1994) illustrate the changing importance between the three main types of processor units (Figure 5.1). This change has been a response to the fact that the performance/cost ratio of personal computers has undergone dramatic improvements, especially during the last five years, and advances in technology have tended to blur the distinction between the levels of processors. A PC today could out-perform most mainframes of 15 years ago, and most workstations of five years ago.
Table5.2 A comparison of processor units (after CCTA, 1993)
|Processor Type||Number of Users||Main Memory||Role||Cost Range (Sterling)|
|Mainframe||More than 256||64MB+||Volume processing||£250K plus|
|Minicomputer||Up to 256||8MB-256MB||File server||£50–£250K|
|Workstation||Single user||4MB–64MB||Stand alone||£3–£50K|
|PC (including lap-tops)||Single user||1MB–16MB||Stand alone or single user node on network||£1–£5K|
Figure 5.1 The Changing Importance of the Three Main Types of Computer Processing Units
For GIS purposes alone a mainframe would seldom be necessary, though some of the more complex packages function optimumly on them. A mainframe would usually only be used if it were already available in an establishment and if the task warranted its use. With the present status of marine fishery resources GIS it is unlikely that there would be many datasets available or tasks which were required which would necessitate access solely to a mainframe. However, with the rapid build up of datasets resulting from oceanographic remote sensing capability, from other oceanographic research activities and from the increasing need for 3-D or 4-D applications, perhaps using acoustically acquired datasets, then the need for future access to a mainframe might be desireable. But whether this need will grow at a faster rate than improvements in the technology or capability of the “smaller” processors it is difficult to forecast. So presently, for most conceivable large applications of a marine fisheries GIS, a minicomputer would provide sufficient functionality. As Table 5.2 infers, minicomputers may be functionally conceived of as small mainframes.
For most serious GIS applications a workstation would be a logical choice, and indeed much of the GIS software is currently supplied to function on specific models. Most workstations have a 32-bit architecture, a large main memory and storage capacity, use the UNIX operating system and would utilize a high resolution graphics screen of typically 21" dimension, often in association with a mono screen to display the current processing status. They typically function as stand alone systems though they may be integrated on to some a network if required (see Section 5.3.2). Frank et al (1991) envisage that by the late 1990's the average GIS workstation will have the following specifications:
* A processing unit having 500 MIPS (million instructions per sec.
* 500 Mb of RAM memory.
* 5 Gigabytes of storage space on internal hard disks.
* An additional 50 gigabytes of memory on external optical disks.
* A workstation screen with a 2 000 by 2 000 pixel display.
* A communication device with 100 megabits per sec transfer rate.
The main growth area in GIS, with regard to the numbers of software products sold, is in the PC (or micro-computer) range. This range incorporates a huge variety of brand name processor products, most of which can be classified as either being PC's or Apple Macintosh's (MAC's). The former usually operates under the MS-DOS system, uses an Intel chip and they are said to be IBM compatible. MAC's use Motorola micro-processors and have their own propriety operating systems, though MAC's which can also run under DOS are now available. In view of the large potential data inputs, and GIS requirements for copious mathematical calculations then, if a micro-computer is being considered for marine fisheries GIS purposes as a stand alone machine, certain minimum size requirements should be recognised. For instance, for a PC at the present time, an 80486DX chip with a speed of 50MHz, at least 8MB of internal memory (RAM), a 200 MB hard drive plus a 16-bit SVGA high resolution card with a suitable colour screen, would all be minimal necessities. Ideally a more powerful processor would be recommended but decisions on this would accord with the scope of the GIS envisaged. Obviously, the purchaser would also need to consider such things as the number of expansion slots, the size of cache, the size of floppy disk ports, serial ports, parallel ports, etc. For most GIS purposes, we would not recommend consideration of lap-top or notebook computers simply because the screen size can only give a limited visual image. Finally, we envisage that there will be a blurring in the distinction between PC's and workstations, i.e. there will be a continuous range of computers from small PC's through to large workstations.
As with other computer technology, but perhaps not so noticeably, Cathode Ray Tube (CRT) display has been advancing rapidly. High resolution displays with a dot pitch of >0.28 mm are now usual for most PC screens, giving a standard resolution of 1024 horizontal dots in 768 vertical lines. With the larger screens on most workstations there may be a 1600 × 1280 screen dot image. Screens now have “non-interlaced scanning” which gives a flicker free display, and screens will commonly have low-radiation, anti-static and anti-magnetic features. The colour capability is now vast such that 24-bit memory allows for the potential display of 16.7 million colours, though the capability of the programme may restrict the colour range. Screens can now be commonly transferred between different computing environments, i.e. since they support an array of graphics adapters (typically IBM MCGA, VGA, 8514A, XGA, Apple Mac II and VESA). Costs vary from about US$1 000 for a 17 inch screen to US$3 000 for a 21 inch screen, though most low cost standard 14 inch modern screens can also be used.
In this section we will not be concerned with the working memory or random access storage (RAM) - we are only concerned with secondary storage. It is useful to note that secondary storage functionally comprises of two activities:
a) The storage of computer programmes, or sometimes datasets, which have usually been purchased and which consist of data which remains unchanged, i.e. read only material.
b) The storage of files which usually consists of work being done by the computer user and which could be constantly changed.
This distinction is important because some storage devices will only cater for (a) above.
Most computer users will be familiar with the fact their data is normally stored on 5.25 or 3.5 inch “floppy” disks or on the processor's internal hard disk drive. Floppy disks have an extremely limited data capacity (1.44 MB for a high density disk), though hard drives for PC's may range upwards from 10 MB to at least 5 GB (5 000 MB). Obviously the storage capacity on larger processors is much higher and the actual amount stored may reflect the specific configuration in use. Storage capacity is required not only as a primary depository of data but also as a back-up for this data.
An external hard drive is sometimes favoured. This may be a simple way of extending the processors capacity, i.e. if the processor has insufficient room for adding extra data storage or as a way of storing data separately from the main computer in case of an computer “crash”. This device often acts as a back-up storage system, i.e. for the duplication of files. Storage costs using a hard disk work out at about US$1.50 per megabyte.
Large volumes of data, usually that used on mini-computers or mainframes, may be stored on a tape drive. These are either magnetic tape cartridges or reel to reel drives. These portable storage methods allow for off-line storage, i.e. when a particular tape is required it must be retrieved and accessed to the processor. Digital Audio Tapes (DAT) and video cartridges are becoming more popular because of their capacities and compactness. Magnetic tape is the cheapest form of digital storage with costs working out at about 5 cents per megabyte.
Another storage medium for accessing large volumes of data are CD-ROM's (compact disks - read only memory). CD's are optical disks which incorporate laser technology, and they are capable of storing huge quantities of data on their three mile long pitted spiral surface, i.e. each disk can store up to 650Mb. This efficient storage medium means that, as their speed of operation is now increasing, thanks to double or quadruple data transfer rate times, then they could become the software source format for all larger programmes in the near future. One of their main advantages is that they are easily archived and transportable. Optical disks which run on the conventional CD drive are now available which can be written to, edited and then re- written. Many micro-computers are now being sold with CD-ROM drives installed and many larger GIS datasets are being released on CD's.
The range of printers and plotters is now very large indeed. Our sub-headings here reflect those which are commonly used, but these output devices could be classified differently. Indeed, in the literature there often seems to be no clear distinction between the term “printer” and “plotter”. Before any decision is made to invest in any of the output hardware described, the final user would need to think carefully about factors such as the speed and throughput of output, the nature of the output involved, special paper and ink requirements, plus the size, type, quality and volume of this. Details on plotter selection are given in Thomas (1993) which also provides details on plotter price and performance (Figure 5.2).
These are low cost, and therefore popular, devices which may be used both for the high volume output of textual documents or for slower output of graphical material including maps. Output is typically at A4 size though many devices can easily produce A3 size or longer output on continuous traction paper. Printing speeds for textual material varies from 300 to 2000 lines per minute depending on the quality of the device (plus its printer settings), the software being used and the quality and size of the fonts being currently used. Graphical output is usually slower. Here, the software organises the output so that it is printed sequentially line by line. Output is via a series of dots with the resolution being limited by the density of pins on the print head - it is typically in the range of 70 to 150 dots per inch. Some of the older graphical packages used to obtain shading density from line printers by varying the combination of alphanumeric letters which they printed. Multiple colour ribbons can be used to produce a limited range of colours. Output from dot-matrix machines is relatively crude, and noisy, but these drawbacks should be measured against cost advantages. Costs per machine vary within the range of US$200 to US$500.
Figure 5.2 Plotter Technology: Price Against Performance
Using laser technology, these printers are capable of very high quality, almost silent output of both textual and graphical information. Until recently most output was limited to black and white shades but colour models, at rapidly falling prices, are now available. The size of machines varies considerably, from low cost A4 models (desk-top or portable) having an output of about 15 pages per minute and a resolution of 300 dots per inch, up to large A0 printers with a variable resolution of from 400 to 600 dots per inch. A4 laser printers can now be purchased for US$600, but laser plotters will be more expensive, e.g. a typical A3 laser plotter would be US$6 000 but an A0 model could be in excess of US$45 000. Output from top end laser plotters would be equal to that achieved from conventional offset printing.
These devices fill a similar niche as the laser printers, with the difference being that they are usually cheaper to purchase being based on a different technology. Output is obtained by forcing ink through small nozzles to form microscopic droplets which strike the printing media in combinations determined by the printing commands. By mixing any combination of four basic colours via multiple nozzles, an almost infinite array of colours can be achieved. The resolution varies from 120 to 300 dots per inch, though the newest A0 ink jet plotters have a resolution of 720 dots per inch. Printing speeds may be comparatively slow but with recent improvements in price, speed, resolution and reliability they are now extremely popular as a GIS output device. In fact, we would envisage that in the immediate future, with the reduction of “fuzzy” outlines, then ink-jet technology will offer the best all-round solution to the GIS user's needs. Small colour desk-top ink-jet printers cost about US$500 whilst the A0 size is about US$6 000.
This type of plotting is basically mimicking what a cartographer or draughtsman would be drawing by hand. There are two basic types of pen plotter - drum (or roller) and flat-bed, each of which has particular advantages. Both types utilise cartridge pens, which can be of variable colours or line width, and they both produce vector (or line) output which corresponds to grid co-ordinates held in the computer file to which they are linked. Maximum line drawing speed is about 100 cms per second and “intelligent” plotters can now arrange vectors in the plot file so as to plot them most efficiently. Some top of the range plotters are also designed to operate as precision scanners with a resolution of up to 400 dpi. A major problem with pen plotting has been the need for constant supervision in case pens clog or run out of ink, but unattended pen plotting technology is likely in the near future.
(a) Drum plotters (Figure 5.3) consist of a drum, which can be rotated in both directions, which has attached a continuous roll of paper or other drawing media. The pen(s) moves horizontally in both directions along a carriage to plot lines. Drum plotters vary in width from 25 to 150 cms, and output can be in up to eight colours. Their main advantage is that they can produce multiple plots without the need to feed paper and the plot length is flexible (up to 50 metres). Cost for a small A4/A3 desk-top pen plotter is about US$1 000 whilst an A0 drum plotter is about US$5 000. The newest desktop pen plotters, and some of the larger free standing models, now use rollers (rather than a drum) to move the paper backwards and forwards.
(b) Flat-bed plotters have a flat drawing surface over which the pen(s) can move in x and y directions or in conjunction to produce curved lines. These plotters vary in size from small A4 to larger A0 and the output is again controlled by co-ordinates held in the host computer file. Their main advantage is that they can utilise a wider variety of media, e.g. paper, plastic, engraving foils, or photographic paper, and scribing can be done by several types of pen plus laser implements. Very low cost products (from about US$300) can now be obtained for attaching to micro-computers, or which have their own drives which allow the user to simply insert his disk and the plotter will then work from the data stored, i.e. freeing the computer for other work. Top of the range A0 flat bed pen plotters cost about US$4 000.
Figure 5.3 Typical Drum Pen Plotter
These devices, of which there are several sub-categories, operate by using an array of nibs (electrodes) to selectively deposit charges on paper (or film) in the form of an image. Toner is then made to adhere to the charged areas, i.e. in a similar way to photocopying machines. Once the data has been converted to a raster format, images can be output from the plotter at 60 × A0 mono drawings per hour, or 15 cms per second for A0 colour output, i.e. much faster than any pen plotter. Output can be in mono or colour, at any size, and resolution is high at up to 400 dpi. For large format output on a daily basis, this type of plotter might be a realistic option. Although costs have been decreasing, these plotters may still be priced upwards from US$40 000.
Direct thermal plotters (Figure 5.4) use local heating to warm thermo-sensitive paper, which is coated with two separate, colourless components. Once heated, these combine to produce a wide range of colours of fair quality. They can produce output, in raster format, up to 15 metres long and have a resolution of over 400 dots per inch. Although they are expensive (about US$12 000 for a A0 wide body colour version) they have several advantages, e.g. no cartridges, toner or ink to consider, no feathering of lines due to ink absorption and they can run totally unattended.
Output rates for a full A0 map can be just 30 seconds, and the quality of this can equal that of photographs. Smaller A4 and A3 colour thermal printers are now available which utilise a similar technology and cost upwards of US$5 000. Figure 5.5 compares approximate total plotting costs for an inkjet plotter (HP DesignJet) compared with a direct thermal plotter (Oce G9000). These lower output costs plus lower operator costs and faster output compared with inkjet plotters must be placed against the higher initial purchase price.
Figure 5.4 Typical AO Thermal Plotter
Having discussed some of the major hardware requirements, we are now in a position to give some ideas on the selection and arrangements of the various hardware components, or the GIS “architecture” or “configurations” as they are sometimes called. We discuss configurations which progress from the simple to the more complex. In doing this there should be no inference that systems progress from “bad” to “good” or from “make do” to “ideal”. Clearly, the choice of systems architecture will be a reasoned judgement reflecting particular circumstances relating to personal preferences, finances available, any functional requirements, the number of users and the required degree of interaction with other systems. Whichever type of configuration chosen, there will always be the possibility of additions to, or upgrading of, the existing facilities. In Chapter 8 we discuss further the task of systems selection.
Figure 5.5 A Comparison of Operating Costs Over Three Years for Inkjet v. Thermal Printing Technology
The minimum hardware configuration can be met by having a processor (PC, Apple Mac or a workstation) linked to an output device, usually a laser or inkjet printer or sometimes a small pen plotter. This configuration could allow for a wide range of output to the screen, for saving to disk or in hardcopy format. However, the input of data to the system would be limited in the sense that it could only consist of tabular data, which may be stored in a spreadsheet or in a database package, or data stored on a floppy or CD-ROM disk which might have been purchased as a ready made data set or had been captured via a data conversion company (or via any other agency). So if the regular interchange of data is envisaged, then stand alone systems are not ideal. These systems are also not ideal if speed is required or if large data sets are to be handled. However, additional functionality can be obtained by adding a digitiser, or a scanner to a stand alone system so that user defined mapping outlines could be captured. For most stand alone systems it would also be a simple task to expand their functionality by establishing communications linkages to external computers or data sources. The cost of a minimal stand alone system, including software and hardware, would be US$3 000 though it could rise to about US$30 000. Figure 5.6 gives an idea of what this configuration could consist of.
This configuration includes a variable range of hardware, usually including a central processor (sometimes called a file server or host computer because part of its function is to facilitate the sharing of the various devices), several PC's or workstations, a printer, a plotter, a scanner and digitising facilities plus probably some external data backup storage device. This system thus becomes a network (usually called a Local Area Network - LAN) and it has distinct advantages, not only in terms of power, but also from the fact that several operations can be performed simultaneously and the expensive peripherals can be conveniently shared (as can any software and indeed any data). The network may also be integrated with other information systems within an organisation. Figure 5.7 gives an indication of a typical departmental system. There are two basic variations of this LAN, each with advantages and disadvantages. The first is where each workstation performs an allocated task having its own data storage and set of software programmes (so-called intelligent workstations), and the second is where all workstations are connected to one main network server on which the software and data is stored. Clearly there are going to be large cost considerations if this type of configuration is installed. These costs are not only for additional hardware, but also the software has to be licenced for each machine on which it is installed, and there are greater needs for database management (see section 6.5).
Figure 5.6 A Basic Stand Alone GIS Configuration
Figure 5.8 gives an indication of the possibilities for operating a marine fisheries GIS on a considerable scale and with processing being distributed over a wide spatial area. This configuration becomes an enlargement of the local area network in the sense that all GIS operations can have access to databases which might reside almost anywhere - so-called distributed computing. For example, larger corporations or perhaps government departments will frequently have databases residing in different locations, and microwave transmissions or cable networks allow transference of data between workstations regardless of their location. The wide area network (WAN) can consist of an almost indefinite array of hardware devices. Clearly the advantages of this configuration are in terms of processing power, the rapid access to datasets which are in scattered locations, and the consequent time savings in assembling requisite GIS information. However, these advantages are only gained at some considerable financial expense in terms of expertise, organisation, capital expenditure and data gathering costs.
Figure 5.7 A Typical Centralised, Departmental GIS Configuration on a Local Area Network
Most workers in GIS research would contend that the configuration future lies in WAN's. The main reason for this lies in the need to access distributed databases. We are already aware that any fisheries GIS is likely to be involved in a range of different and diverse subject areas (further discussed in Chapter 7), and thus eventually there are going to have to be linkages to other departments or establishments. Table 5.3 gives some reasons why distributed processing via WAN's is advantageous. Laurini (1994) gives an interesting account of some of the considerations in the move towards distributed database use in GIS.
Table5.3 The Advantages of Using Distributed Computer Capabilities for GIS Work
|*||Cost and effort savings on the need to duplicate databases.|
|*||Far less total data storage capacity is required.|
|*||Any one database maintains its integrity since it will be in the interests of the holder to maintain and update it regularly.|
|*||Easier for users not having to maintain security over their databases.|
|*||Data in effect becomes both local and international.|
|*||Access to distant data becomes easy.|
Since there are no logical ways of examining GIS software then, after a brief introduction, this section will seek only to (a) discuss some of the varied characteristics of the software in general and (b) examine in some detail a range of specific GIS packages. It is important to point out initially that we will not be looking at types of software that are ancillary to GIS yet which, in some cases, may be vital to its success, e.g. we will not consider any operating systems, any image processing packages, any graphical enhancement or drawing packages or any specific database management programmes (though the latter will be mentioned in Chapter 6). It is also important to note that there is no exact definition of what GIS software is! With this in mind we will also not be including here considerations relative to:
Figure 5.8 A Wide Area Networked GIS Configuration
* Database management systems (DBMS).
* Cartographic (mapping).
* Computer Aided Design (CAD).
* Remote sensing or image processing.
* Digitising programmes.
* Contouring and surface modelling.
i.e. even though any of these elements might be incorporated within many GIS software package. Some points to consider when purchasing software are considered in section 184.108.40.206. Until the late 1980's, one of the main criteria upon which the choice of GIS software was made, was whether the system was raster or vector based. Until then it would be true to say that vector systems, though often more expensive would generally have been preferred, i.e. given the types of use for which most earlier GIS's were acquired, plus their ability to provide a better looking final output and the lack of affordable raster data or methods of data capture. However, since then the balance has slowly shifted in favour of raster based systems. This is because:
(i) New computer processing capacity has allowed for easier storing of large raster data quantities.
(ii) VDU's can now display huge arrays of very small pixels, i.e. providing a vary clear eight-bit colour image.
(iii) The resolution of raster scanning devices has dramatically improved, i.e. they are now able to scan at 1000 dpi. Once scanning is above 500 dpi it is difficult to differentiate between raster or vector lines.
(iv) The quality of scanned raster data, using various collection techniques, has greatly improved.
(v) Larger amounts of commercially available raster data are now both available and affordable.
(vi) Progress has been made in raster processing techniques which allow for automatic vectorization.
(vii) Advances are being made in introducing effective standards for raster data handling, interchange and storage.
(viii) Raster images are now frequently being used as background drapes on vector images, i.e. in order to aid visual interpretation of user data.
Since software characteristics are so varied, we can only attempt here to offer some insight into a few of the ways in which packages differ. It is important to know something of this because, given the range of GIS software which is available, it will pay the prospective purchaser to be familiar with some of the differentiating factors which need to be borne in mind before investing in a particular software system.
GIS applications software consists of multiple programmes which are integrated so as to ensure the capabilities of mapping, managing and analysing geographic data. The balance of the programmes in the package may be biased towards the database management side or towards graphical presentation, etc, i.e. according to the particular market niche which the software is aimed at. There are also specialised packages which may be marginal to GIS which highlight certain areas such as network analysis or terrain analysis. Basically the GIS software should allow for a minimum of:
(a) Graphical data entry - allows the inputs of map features in the form of geo-referenced locations.
(b) Annotation entry - allows for textual information to be displayed on maps.
(c) Graphic editing - allows maps to be updated or amended.
(d) Data manipulation - allows the data to be manipulated in some or all of the ways outlined in Section 6.2.
(e) Graphic display - allows for control over the appearance and format of maps.
(f) Database management - allows for the entry, storage, retrieval and management of information relative to the graphical entries.
The software must have a language by which queries are made (usually Standard Query Language - SQL), and some form of user interface. The software should also support a wide range of peripherals through having the relevant drivers, i.e. the routine that handles the specific details and characteristics of a single peripheral device so that it can work correctly. Most software is now produced to meet with official or de facto standards, e.g. most important for GIS software is ISO 8211. These standards have been adopted to regulate the way in which the broad GIS community of vendors and users operates
The first specifically GIS packages were produced in North America during the late 1970's. Their evolution and categorisation has been complex. One of the main reasons for this is that it is still difficult to accurately define what is, or is not, GIS software. For instance, there are many mapping packages available which have varying degrees of functionality and obviously some of these packages will offer peripheral or part GIS capability. There are possibly 60 to 80 true GIS packages which are now commercially available; indeed Cassettari (1993) lists 42 GIS packages which are available for microcomputers. There is also a large amount of GIS software which was originally designed for a particular purpose, but which has later spawned a new and similar commercial package, e.g. if one electric utility company acquires a purpose built GIS then this product is often suitable for other electric utilities.
Much of the software capability has evolved from Computer Aided Drawing (CAD) packages in conjunction with the field of computer graphics. The RS industry has more recently given the incentive to integrate much image processing capability into GIS software. And most recently, as has been argued by Antenucci (1993), it is database capabilities which are the driving force behind GIS software evolution. Whereas once software was either dominated by a raster or vector data storage considerations, packages are increasingly able to cope with both of these. The market for GIS software has been growing worldwide at about 15% per annum for at least the last five years.
It has been argued by Frank et al (1991) that GIS software development is a major problem and that its rate of development has been considerably slower than that for hardware. This has led to both an increase in prices and a poor record of reliability including on time product development. The authors suggest that, because of the poor product development record, many of the software sub-systems, e.g. programming languages, database management systems plus operating systems used, are quite dated. It is actually unlikely that most users would have noticed this “software crisis”, i.e. since new software versions are always entering the market, since user interfaces have greatly improved and because many users are aware of software developments which are taking place such as research into and implementation of object oriented database management systems
Most of the GIS software has originated from the USA. Here there are two main classes of legal ownership, i.e. that in the public domain (government funded and researched) and private ownership. This has caused major problems in the past since public domain software has been sold at prices which cover little more than marketing and distribution costs, and since these products have sometimes been acquired by private companies for integration into their packages. This problem is now diminishing rapidly as commercial software prices have generally fallen and as public domain suppliers have sought to make greater financial gains from their products. There are now about half a dozen major private software houses, which have grown rapidly in size through both sales growth and company mergers, plus dozens of minor more specialist companies. The two largest GIS software suppliers on the world market are ESRI and Intergraph, who between them have 44% of this market (Frost & Sullivan, 1994). Some of the main suppliers are also the public domain companies who exist in Europe as well as in the USA, e.g. ILWIS in the Netherlands and GRASS and IDRISI in the USA.
It is difficult to describe the software range since it is so vast. Products which are purely raster based tend to be cheaper since it is far easier to programme the software. This means that lower end raster products which have emerged from public domain sources can still be purchased from about US$150, though prices vary according to the status of the purchaser. At the other end of the “off the shelf” market, there is software which is designed for use on mini or mainframe computers, or high end workstations, using the UNIX operating system and which has an extremely wide functional range working in both a raster and vector format. Prices here might be in the region of US$10 000. It is fair to say that most software has been designed with a bias towards certain types of operation. Some examples would be:
(a) Products that accentuate their RS integration capabilities.
(b) Products which are particularly suited to environmental work.
(c) Products which are described as “desk-top mapping” packages.
(d) Products which are best suited to GIS training or research.
Certainly, no two GIS software packages would be anything like one another and this makes comparisons very difficult. Additionally, as hinted above, many of the larger GIS's are specifically designed for individual purposes. A recent trend being followed by many of the suppliers is to offer their software in the form of a number of separate modules. This partly results from the fact that the software was becoming too complex, i.e. offering an incredible breadth of functionality, and most users do not require this, although it is also a way of selling a wider range of products.
The user interface is important in that it determines how effectively the user can work with the system. Thus it should hide internal details so that the GIS tasks become the focus for attention. User friendliness varies quite substantially between packages, i.e. though most now have menu or icon systems, some still require a complex command language structure to be learned. The learning process for some systems may take many months, though this depends largely upon whether learning takes place via gradual familiarization with use or via attendance at a specific GIS training course. Unfortunately, some of the menu driven interfaces are still very complex so learning is still slow. This is often because insufficient thought has been given as to how the typical GIS user thinks, and thus how individual menu items should be grouped. What needs to be done in a computing situation as complex as that of many GIS packages is to design the user interface first, and then integrate it into the software. Raper and Bundock (1993) give an excellent account of the variability of graphical user interfaces, and the importance of developing GUI's which are both easier to use and able to be user customised.
These can be most usefully listed:
(a) Some have internal database management systems whilst others do not.
(b) Their ability to service a range of hardware devices varies.
(c) The range of product support and training varies greatly between suppliers.
(d) Some systems require the use of two screens (VDU's), one for the graphics and one for the textual command information.
(e) Some systems can accept user programme modules as well as those that already integrated into the software.
(f) Some systems have windowing capabilities which allow for two or more areas of interest to be displayed on the screen at one time.
We have chosen five GIS software packages as examples. Our selection was simply on the basis of exemplifying a range of packages in order to give the reader some familiarity, not so much with what they can each do, but more with how much they vary from one another, i.e. in order to get some appreciation of the care necessary in selecting a GIS. One of the packages chosen, Intergraph's I/SEA VUE, is not a full GIS but it has been included as an example of the many specialist packages which are becoming available which can sit comfortably alongside a marine fisheries GIS.
ARC/INFO originated in 1982 and is now one of a range of GIS products produced by the Californian company Environmental Systems Research Institute (ESRI). It is a “top end” product which was designed to function on mainframes, mini and workstation computers but which can now also be used on PC's or indeed in a mixed machine environment consisting of a network of virtually any combination of GIS related hardware devices. The system originated as a pure GIS tool with the ARC software component being used to manage cartographic data and INFO used to manage the tabular attribute data. ARC/INFO has an extensive functional capability including some 2 500 geo-processing tools. It can support both centralised and distributed databases and its open architecture means that it can be readily linked to other software. The latest versions have a convenient GUI, via a suite of menus, called ArcTools. This was very necessary since originally ESRI were comparatively slow to adopt a simple GUI, which meant that a lot of prospective customers were deterred by the complex command language which formerly had to be learnt.
As with other major GIS suppliers, ESRI have now modularised ARC/INFO so that its GIS capabilities are available via a number of smaller, more specialised packages such as ARC/Cad and ARC/View, all of which share similar database handling capabilities. The company are also offering ARC/INFO as part of a number of integrated turnkey packages. This means that the certain of the software components are combined, together with the requisite hardware, to form a specialised total package for such applications as land and property management, highway information systems, map management and production or various census applications. Many of these specialised application areas represent those which would have similar types of data inputs, e.g. inputs for a census application would all be in the form of tabular numeric data which was geo-referenced to varying levels of the census enumeration district hierarchy, i.e. in terms of most of the mapping, this would be simply relating numbers to unit spatial areas. A further move by ARC/INFO is towards the greater demands for spatial analysis and spatial decision support tools rather than GIS applications which were based on simple inventory monitoring and reporting. New versions of ARC/INFO come out almost every year, and ESRI hold an annual users conference in Palm Springs, California. Prices for ARC/INFO vary enormously according to what you buy and who you are - for a main frame version they might be US$15 000 for the complete GIS, but a PC version for an educational establishment the price is about US$2 500. Figure 5.9 gives an example of the output from ARC/INFO as applied in a marine environmental context.
Atlas GIS was developed specifically as a PC based vector mapping product. It is a top of the range GIS having its own internal database management system, and its makers (Strategic Mapping) have deliberately oriented it towards the business community as a “desk top mapping” tool. Thus it is an off-the-shelf product which comes “bundled” with various “free” outline datasets which are either European or North American based. It is anticipated that many businesses will have either their own customer databases or be able to purchase requisite data sets containing some form of geo-reference (frequently post-codes). These can be simply integrated into Atlas GIS, or the data can be entered via a Lotus type spread sheet, for subsequent mapping. The makers claim that “Atlas GIS for Windows is a way of making GIS a fully integrated tool of the computer desktop”.
Figure 5.9 Example of ARC/INFO Output Showing Marine Sediments in Parts of the Wadden Sea, Germany (from Liebig,1994b)
Atlas GIS has a clear user friendly interface allowing maps to be easily constructed. By clicking on the various menu bars, full menus or windows are opened up, most of which are easy to follow. Figure 5.10 provides a screen image with an indication of some of the ATLAS GIS functionality. Screens can be fully customised. Most mapping output is via density or colour shading, though dot density maps or proportional symbols can be drawn. Maps can be copied to a clipboard and pasted to other Windows applications. Up to 250 map layers can be drawn. One of the drawbacks to Atlas GIS is that it does not support digitising, though there is a DOS version which does. The software runs on an IBM-compatible 80386 PC though a 80486 is recommended (as are 8MB of RAM). The price for Atlas GIS for Windows is about US$2 300.
Figure 5.10 Screen Image of ATLAS GIS for Windows Showing Some Main Features
IDRISI is a raster based geographic analysis and image processing system, developed by the Graduate School of Geography at Clark University in the US. It is designed to provide a professional level research tool on a low cost, non-profit basis. Since its introduction in 1987 IDRISI has grown to become one of the largest raster based microcomputer GIS and image processing systems on the market, i.e. accounting for 40% of raster systems worldwide. IDRISI is used for a huge variety of tasks including many under the United Nations Environmental Program (UNEP) GRID programme.
IDRISI is a collection of over 100 programme modules. The modules are linked by a uniform menu system. The modules fall into three broad categories:
(a) Core modules - these provide the fundamental facility for data entry, storage, management and the display of raster images.
(b) Analytical modules - these consist of the major tools for the analysis of the raster data. These modules can be divided into three main groups - geographic analysis, statistical analysis and image processing.
(c) Peripheral modules - these are associated with data conversion between IDRISI and other programmes, data import and export and data storage formats.
By using this modular structure it allows users to develop their own analytical modules which can be linked with programmes in the IDRISI core module. There is also a new interactive digitising module (TOSCA) which supports any digitizer capable of ASCII output. IDRISI is able to run under DOS on almost the minimum PC with only 512K of RAM, a small hard drive and colour graphics card. In late 1994 a Windows version of IDRISI was released, which among many features included a simple means of translating the software into different languages and alphabets. IDRISI can produce raster images of up to 32 000 rows by 32 000 columns and has full vector to raster conversion. It also has import/export facilities to all the major GIS's. This large range of functionality can be purchased at about US$650 for commercial buyers but only US$150 for students. IDRISI also operate a bulletin board which is connectable via the E-mail address:
and they are in the process of setting up a number of IDRISI Resource Centers throughout the world which, operating from local university sites, will be able to offer customers direct training, advice, materials and other services.
An example of output is shown in Figure 5.11 which illustrates high resolution raster output from part of the tutorial package which comes with IDRISI. IDRISI for Windows software system is now available.
ERDAS Imagine is a GIS which has grown from an original image processing package, and it is one of several such packages which ERDAS supply. It is a highly sophisticated raster based product which not only performs all standard image processing and photogrammetric requirements, but which also allows for the overlay of vector data so that both vector and raster analysis, as well as other spatial modelling and map compilation procedures, can be performed. The raster images can be satellite, aerial, photographs or in virtually any form. The GIS has an icon based GUI and it runs on UNIX based systems or in Microsoft Windows NT. ERDAS Imagine consists of six core modules, plus other add on modules. The core modules are:
Figure 5.11 High Resolution Raster Output From IDRISI as Used for Pattern Recognition Training
(a) Standard - this is the core of the GIS providing all the input/output capabilities.
(b) Spatial Modeler - this allows raster, vector and tabular data to be integrated, queried and then modelled.
(c) Digital Orthos - this provides photogrammmetric capability allowing for ortho-imagery plus the creation of digital elevation models.
(d) Radar - allows the most common radar datasets to be processed, enhanced and integrated with all other Imagine modules.
(e) Vista - provides access to raster, vector and tabular datasets for querying, processing and composing outputs from anywhere on the network.
(f) Vector - allow for direct access to the ARC/INFO database handling facilities.
ERDAS has been used in a variety of different fields including land classification and land use modelling, vegetation and forest mapping, environmental impact assessment and oceanographic and other marine studies. It is a product which has been used extensively both in the commercial and public domains, including the UNEP GRID environmental monitoring programme based in Nairobi. ERDAS Imagine is a “top end” product whose price would be in the US$7 000 to US$10 000 range depending exactly on what modules are purchased. Educational establishments can obtain substantial discounts. ERDAS offers other extensive “peripheral” advantages and guidance such as a regular magazine, an educational center in the US and video training packages. Figure 5.12 gives example output of coastal zone monitoring being undertaken in Georgia, USA.
Figure 5.12 Typical Output from ERDAS Showing Monitoring Processes on the Coastline in Georgia, USA
This is a specialist software package for oceanographic and nautical mapping. The software allows for the simple plotting of acoustic echo sounding data in map form. Maps can be presented in either contour or 3-D representations. Since echo sounding techniques produce vast quantities of closely spaced digital data, the software is able to automatically thin out the data while still retaining all critical information. The software has additional analytical capabilities which allows it to function as a nautical mapping GIS, though the data can be exported to a full GIS for greater functionality. There is a simpler version of the package called I/SEA which is designed to just remove surplus sounding data, thereby creating databases which contain only the soundings for safe navigation.